Several members have asked about the appropriate dose of avastin (bevacizumab), which is really still a controversial subject. It's worth exploring how we got here and where we are now.

While other doses of avastin have been used with other tumor types, the first study in lung cancer that used avastin tested two different doses, 7.5 mg/kg or 15 mg/kg combined with carboplatin and taxol (paclitaxel). This work was done at Vanderbilt Univ. Cancer Center by Dr. David Johnson and colleagues, and Dr. Laskin worked with them for a couple of years. This study had the following design:

(Click on image to enlarge)

The final report of this rather small trial (abstract here), with fewer than 35 patients per arm, showed better results for the higher dose than for the lower dose, but that may have been because there were more patients with squamous NSCLC who happened to be assigned to the low dose arm. Although this early trial enrolled patients with all NSCLC subtypes, it was this work that first indicated a disturbingly high association of fatal or near-fatal bleeding with squamous tumors.

For the follow-up study that excluded patients with squamous cancers, called ECOG 4599, the design was on a larger scale with nearly 900 patients who were randomized to just two arms: chemo with or without avastin, only at the higher dose that was selected based on looking better in the earlier trial.

As I've described in one of my earliest posts (here), the results were positive (abstract here), demonstrating a significant improvement in the survival of the recipients of chemo and avastin, and this led to the FDA approval of avastin for lung cancer, at the higher 15 mg/kg IV every three week dose, with the carbo/taxol regimen.

After the presentation of this trial, and certainly the FDA approval, this dose and schedule became a pretty clear standard, except that oncologists who also treated colon cancer were dubious why a lower dose appeared to be better in that setting, but we used twice the dose at twice the expense and also potentially a greater risk of side effects in lung cancer. This issue would be addressed by a trial called AVAiL (AVAstin in Lung Cancer), which had a similar design to the US-based study but used a different chemo of cisplatin/gemcitabine and included three arms: one for chemo alone, one for chemo with low-dose (7.5 mg/kg) avastin, and a third with the higher (15 mg/kg) dose:

The world has been anxious to get answers about the relative value of the two different doses, although technically the trial was designed to test the value of each dose only against their placebo arm and not against each other. Still, it's hard not to compare the two arms against each other, especially if I show the curves for progression-free survival (PFS) for the two doses vs. placebo side by side:

Now while the trial wasn't designed to compare the two doses against each other, to me it appears that the curves have a more compelling separation at the lower dose (left). Importantly, however, the presentation of these preliminary data at ASCO 2007 (abstract here) did not have overall survival data, which were not mature enough to present. Also importantly, there were not clear differences in side effects to indicate a deleterious effect from increased toxicity with the higher dose -- just a slightly increased proportion of patients with moderate to severe high blood pressure. But no increased bleeding or other dreaded side effects at the higher, FDA-approved dose.

Following that presentation, lung cancer expert Tom Lynch from Massachusetts General Hospital in Boston provide his commentary and argued that the lower dose was now a strong consideration as the appropriate dose to use in lung cancer, since it would be nearly impossible for a PFS curve that looks convincingly better at the low dose to be eclipsed by the higher dose in terms of overall survival (OS) in the future. Since then, several experts have argued rather forcefully that the higher dose is the only one we have OS data with, so the ECOG data with high dose avastin trump the half-cooked data from AVAiL that suggest the lower dose is equally good or better.

It's fair to consider this latter argument the "party line" about avastin in lung cancer: most experts favor 15 mg/kg based on the fact that the data with the best quality, and the least controversial endpoint of OS, used the higher dose. While I do still tend to use that dose, it's largely because I feel compelled to adhere very closely to the approved indication: I almost always just use carbo/taxol with avastin, don't give it to patients with squamous NSCLC, brain metastases, history of coughing up blood, or on therapeutic doses of blood thinners, etc. I'm also impressed that if the side effect profile doesn't appear to be convincingly worse with the higher dose, the main advantage of the loer dose is half of the cost (bringing it from wildly expensive to just painfully expensive). However, I must admit that I have a nagging concern about the wider gap in the lower dose curve on AVAiL, that I certainly don't think it's a mistake to use the lower dose based on what we've seen thus far, and that I wouldn't be surprised if the lower dose is the same or better. But I wouldn't expect Genentech, the company that makes and sells avastin, to go far out of their way to finance studies that would lead to use of half as much of their product in lung cancer. Insurance companies or the federal government would need to coordinate this type of work, and I don't see them being proactive or capable enough to ask those kind of questions. So we may end up with further controversy about this in the coming years. In the meantime, it doesn't appear to me that a lower dose is a compromise.